Second, to dye a piece of cloth one mixed shade and by topping with other colors, to alter that shade to match some shade previously selected. For instance, dye a piece a good shade of reddish or copper brown, and then try to “kill” the red in it without materially deepening the shade, i.e., change it from a copper brown to a greyish or dirt brown of about the same depth of color. Very pretty and instructive experiments can be made along this line of building up soft grey shades, by dyeing the cloth successively in weak baths of the three primary colors. As fast as one color predominates, it can be killed by dipping into successive baths of the other two. Attractive scarfs and table covers can be made with a little care, by knotting the material and dyeing light rainbow shades of the three colors, one after the other, changing the knots or tied portions after each bath. Properly done, this will produce remarkably interesting, opalescent effects, each color being toned and softened by the other two, although predominating in different parts of the material. When, in the operation of rainbow dyeing, strongly contrasting colors have been used with unhappy results (such as the red, yellow, and blue tri-color effects that some students will produce) try the effects of toning, or “covering,” as it is often called, with some soft, neutral color which combines in itself all the contrasting tones, or else with a color that is complementary to the most obnoxious one, softening that one and strengthening the weaker shades. Grey, of course, can be used for this; but in general, a soft shade of brown will be found very valuable for taking the edge off of too violent contrasts. The permanganate brown (Manganese bronze), described in the first chapter, can be used with advantage for this purpose. It is not difficult for a skilful dyer to match any desired shade by using three complementary colors, red, blue, and yellow, provided, of course, that these are pure and unmixed. It often happens, however, that after matching carefully a soft mixed shade by daylight, the colors appear entirely different when viewed by artificial light, and especially by ordinary gaslight. Daylight, as we are accustomed to it, is comparatively evenly balanced in color, is in fact a white light. But artificial light as a rule is distinctly colored, and it is difficult, though now not impossible, to find a light that so closely resembles daylight that colors can be matched at night. If the light, for instance, has a bluish tinge, like some kinds of electric light, it will kill the corresponding orange in a shade, while yellow light, such as commonly results from the use of oil, candles, or gas (less marked when incandescent mantles are used), dulls and even blackens lavender, violet, and purple shades, while having little or no effect upon yellow, orange, and green.